Pinhole SPECT Reconstruction Using Blobs and Resolution Recovery

We study the performance of an iterative reconstruction algorithm for single pinhole SPECT, in which the correction for pinhole blurring is realized by averaging multiple projection rays by means of a Gaussian quadrature method. In addition we compare the regularizing effect obtained by using blobs basis functions and voxels with and without postsmoothing. Contrast-noise and resolution-noise studies with a mathematical phantom show that blobs improve the tradeoff between bias and variance compared to voxels with the same grid spacing. The performance of the voxel reconstruction is improved significantly by using postsmoothing with a Gaussian kernel and by taking finer voxel sampling, but the results with blobs remain better provided the parameters of the blobs are carefully selected. We have also reconstructed measured pinhole data for a resolution phantom and for the bone scan of a rat. With these data the visual difference between the postsmoothed voxel reconstruction and the blob reconstruction is small